Food Research International 44 (2011) 331–337

Contents lists available at ScienceDirect

Food Research International
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / f o o d r e s

Nondestructive assessment of freshness in packaged sliced chicken breasts using
SW-NIR spectroscopy
Raúl Grau a,⁎, Antonio J. Sánchez b, Joel Girón a, Eugenio Iborra b, Ana Fuentes a, Jose M. Barat a
Departamento de Tecnología de Alimentos. Universidad Politécnica Valencia, Spain
Departamento de Ingeniería de Sistemas y Automática, Universidad Politécnica de Valencia, Spain

a r t i c l e i n f o a b s t r a c t

Article history: A technique was developed to predict the freshness of packaged sliced chicken breast employing a
Received 2 June 2010 nondestructive visible and short-wavelength near-infrared (SW-NIR) spectroscopy method. Spectra were
Accepted 8 October 2010 recorded at 0, 7 and 14 days using a camera, spectral filter (400–1000 nm) and a halogen flood lighting system
which were developed and calibrated for the purpose. Physicochemical, biochemical and microbiological
Keywords: properties such as moisture (xw), water activity (aw), pH, total volatile basic nitrogen (TVB-N), ATP
breakdown compounds (K1 values) and mesophilic bacteria (cfu g− 1) were determined to predict freshness
degradation. The spectra obtained were related to the storage time of the samples. The best wavelengths for
Chicken meat modeling freshness were 413, 426, 449, 460, 473, 480, 499, 638, 942, 946, 967, 970 and 982 nm. A linear
Packaging correlation was found between the visible and SW-NIR spectroscopy and parameters such as microbiological
counts, K1 and T-VBN indexes.
© 2010 Elsevier Ltd. All rights reserved.

1. Introduction developed based on the measurement of postmortem deteriorative
changes associated with sensory quality, microbial growth (Barat
Consumption of chicken has increased in many countries as it is et al., 2008; Kaneki et al., 2004) and chemical changes, such as pH or
related to a more health conscious diet and mainly due to it being a total volatile basic nitrogen (TVB-N). The concentration of ATP and its
relatively inexpensive protein source, which is ideally suited to the degradation products (inosine-5′-monophosphate (IMP), inosine
many forms of convenience foods in the modern western diet (Ellis & (Ino), and hipoxantine (Hx)) has been widely used in fish freshness
Goodacre, 2001). control (Barat et al., 2008; Fernández-Segovia, Escriche, & Serra, 2008;
In the meat sector, as well as in other food sectors, there is a demand Karube, Matsuoka, Suzuki, Watanabe, & Toyama, 1984; Surette, Gill, &
for fresh and ready to use products, or at least those easy to prepare, Leblanc, 1988). Watanabe, Tsuneishi, and Takimoto (1989) stated that
which explains the popularity of packaged fresh chicken fillets. The the analysis of ATP and its breakdown products could be used as an
booming demand for sliced packaged chicken requires the intensive indication of meat freshness. However, the determination of these
handling of the product to give it the necessary format and causes a compounds in meat has been performed mainly to correlate them
certain loss of freshness. During processing (quartering, skinning, with sensory qualities or with different postmortem meat quality, or
cutting and boning), bacteria on the skin and in the carcass cavity of to test the effect of aw and temperature on the IMP degradation
the chicken are the main cause of spoilage (Alain & Stéphane, 2008). (Flores, Armero, Aristoy, & Toldrá, 1999; Kavitha & Modi, 2007; Mateo,
Food safety is a main concern for consumers. Present economic Domínguez, Aguirrezábal, & Zumalacárregui, 1996). Among the ATP
trends lead to increasing distances between consumption and produc- derivatives, IMP is predominant in meat extract 24 h after slaughter.
tions zones and consequently to an extension of the delivery chain. This compound is gradually transformed into inosine and hypoxan-
Considering these changes, it is necessary to set out methods for thine in the meat (Watanabe et al., 1989).
objective reliable safety control, guaranteeing quality and freshness at Sensory evaluation is, in most cases, subjective and costly.
all stages of the commodity chain (Dufour, Frencia, & Kanea, 2003). Microbial methods estimating bacterial spoilage (Dalgaard, 1995)
Freshness is a rather ambiguous term when referring to meat and and chemical methods encounter problems in measuring early
fish products. Currently there is no methodology to determine meat postmortem deterioration, so that rapid and simple methods for
freshness, but there are non-legislated tests that provide guidance on meat freshness estimation are required (In-Seon, Yong-Jin, & Namsoo,
the subject. Various methods for assessing meat freshness have been 2000; Kaneki et al., 2004; Zhang, Jin, Dong-hui, & Yu-bin, 2008).
Spectroscopic methods have gained importance in the evaluation
of food quality attributes during the last decades (Nádai, 1983; Nádai
⁎ Corresponding author. Tel.: + 34 96 3877362; fax: + 34 96 3877369. & Mihályi-Kengyel, 1984). The fact that NIR spectra reflect several
E-mail address: (R. Grau). parameters of the material suits the method for evaluating complex

0963-9969/$ – see front matter © 2010 Elsevier Ltd. All rights reserved.

/ Food Research International 44 (2011) 331–337 quality (Williams & Norris. Barzaghi. USA). Diffuse reflectance spectroscopy has also been used to A total of 2 extracts were obtained per sample. As they have high .000 ×g for 10 min. Various near. Short-wave near-infrared (NIR) according to the method described by Barat et al. Hx. The Burns and Kee (1985). S. Broadhurst. evaluate the quality of frozen minced red hake (Pink. working at a temperature of 25 °C.1 (Jasco). biological materials (Wu. thus changing their structure and the format in which the 1510). Total Volatile Basic Nitrogen (TVB-N) was determined 27 98 2 according to the procedures described by Malle and Tao (1987). placed in Petri unit (model DG-1580-54). Louis. (2008). Tomelloso. replicate commercial slices. MO. were assayed by HPLC water for Ino and IMP analysis. Naczk. Grau et al. Fernández-Segovia. & Cattaneo. detection and quantification of microbial spoilage of chicken meat (Ellis. inosine. equipment. Nucleotides analysis hydrogen phosphate buffer. The time points for the evaluation were 0 days (immediately after Compounds were identified using retention time comparison of packaging) and after 7 and 14 days of storage at 4 °C. Injection volume was 20 μL. the short-wave NIR spectra are being applied to construct excellent detectors for the The method used to the extraction was similar to that described by nondestructive component determination in biological materials. short-wave NIR spectra can be Barcelona. Spain). were used for microbial and nucleotides monitoring at days 3. Giardina. & Pink. Three slices per chicken breast sample were handled phase was 0. mobile phase A was 0.. solutions of 10 g of homogenized chicken meat and 90 ml of Time (min) Mobile phase A (%) Mobile phase B (%) distilled water were measured with a portable pH meter MM40 0 98 2 (Crison Instruments S. infrared spectroscopic methods have also been published on Barcelona. using calibration curves of peak area of compound vs and nucleotides monitoring). and Hx standards were obtained samples before and after packaging and before and after opening from Sigma-Aldrich (St. S.Extraction of nucleotides reflectance ability in the region of 780–1040 nm. Additionally. mobile slaughter. and Hx were quantified according to the external standard samples were also used for the subsequent analyses (physicochemical method. into a sample with much less heating effect than the long-wave NIR and homogenized for 4 min in a masticator (IUL Masticator. neutralized with solid potassium carbonate and left to stand in ice for 2001)..403. in most of AS-1555-10). and diode array detector (model MD- dishes.085% of TFA were obtained directly from a local poultry processing plant 24 h after in 60% acetonitrile in water. 5 min. auto-sampler (model Sinelli. order to take samples for the microbiological analysis. Data acquisition was performed with ChromPass software meat is marketed. & Goodacre.000 ×g for detectors (silicon diode array) (Mayes & Callis. The chromatograms film (n = 45). For the Ino and IMP analyses. Alella. Spain). The pieces. without fat and skin tissue. inosine (Ino) and hipoxantine (Hx).1. Ino. and 0. internal particle diameter of 5 μm (SW-NIR) as a nondestructive method for determining the freshness (Análisis Vínicos. Barcelona. 2002. Heia & Sigernes. 100 °C (UNE –EN ISO 1442:1979). Italy) with pump (model PU-1580).085% of TFA in 60% acetonitrile in water for Hx analysis. Ino. The aim of this study was therefore to develop and version 17.5. He. mobile phase B was 0.5. evaluation of freshness (Nilsen. & Feng. . For the determination of pH. 2002) or detection of fish spoilage (Lin. K1 values were calculated according to the Eq. pH 4. During the shelf-life study. ternary gradient unit (model LG-1580-02). The entire extraction procedure was performed short-wave NIR region allows NIR energy to penetrate more deeply at 4 °C. degasser these studies samples were manipulated (minced. Lin et al. Spain) following the 15 93 7 procedure proposed by Fuentes. 2006). Analytical determinations Moisture (xw) was measured by oven drying to constant weight at where IMP is inosine 5′-monophosphate. the quality of the broiler chicken cuts . Cavinato & The supernatant was stored at −20 °C until analysis. unknowns with those of standards and by standard addition or “spiking” Image analysis (visual and SW-NIR) was carried out at day 0 in all (Johnson & Stevenson.6 M HClO4 neutralized with solid potassium carbonate. 9 and 12. Al-Holy. Rasco. The unpackaged IMP.2. region (1100–2500 nm). Opening was done in aseptic conditions in prepared in 0. To analyze Hx.Identification and quantification was evaluated as a function of film packaging and storage time. Spain).1. Martens.. the same C18 packing as above was positioned in front of the analytical column to protect it from contamination. Five grams of muscle was mixed with 50 mL of 0.332 R. Kell.9 mL/min in all cases. (1): 2. 2005). IMP. Destructive analyses Ino + Hx K1 ð%Þ = × 100 ð1Þ IMP + Ino + Hx 2. to phase A was 0. 2001). consisting of inosine-5′-monophosphate B: 0. The aseptically and packed randomly in a plastic tray sealed with plastic elution program in both cases is shown in Table 1. prediction of sensory quality criteria (Warm. . for Ino and IMP analysis.HPLC analysis 1999). 1989). 1978). the 15 remaining samples concentration of compound. pH 4. 2008). Standard solutions were (n = 10) at 7 and 14 days. The contents of the stomacher bag were immedi- measured with inexpensive light sources (tungsten lamps) and ately transferred to a centrifuge tube and centrifuged at 12. Barat.01 M dipotassium 2. However.A. with some spectroscopy shows promise for fast and nondestructive analysis of minor modifications. (Milano.1. The supernatant was filtered through glass wool.6 mm. into slices of 200 g each with a thickness mobile phase B was 60% acetonitrile in water. and Serra 16 0 100 26 0 100 (2008). Also. Esaiassen. were cut. which trifluoroacetic acid (TFA) in water.1% of The study was carried out using 50 broiler chicken breasts. Water activity (aw) was analyzed with an Aqualab GB-X Fast-Lab (GBX. Rowland.01 M dipotassium hydrogen phosphate buffer. The neutral extract was centrifuged at 12. Mousavi. and 60% acetonitrile in (IMP).1. under identical chromatographic conditions. 42 98 2 A: 0. France) Table 1 Elution program for HPLC analysis. Ino.1% of trifluoroacetic acid (TFA) in water for Hx analysis. Flow rate of mobile of 20 mm. A guard column containing of sliced and commercially packaged chicken breasts. 2. 2006. The analysis was conducted on a liquid chromatograph Jasco 2004) and monitoring the shelf-life of dairy products (Sinelli. 10 min in a refrigerated Medifriger BL centrifuge (JP Selecta. Romans-sur-Isère. A.L. & Nielsen.6 M HClO4. Separations were achieved on a reverse- evaluate the feasibility of visible short-wavelength near-infrared phase Ultrabase C18 250 × 4. Materials and methods The composition of mobile phases was different for Hx and for both Ino and IMP analyses. were monitored at a wavelength of 254 nm. etc). The ATP-related compounds. hypoxanthine.

homogenizing of samples. The scanned line is composed of 1392 points. Aerobic counts multiple regression analysis and the forward selection method were were determined by using plate count agar (Merck. with a distance of 40 cm between the camera and the sample. The illuminants were two halogen The relationship between the spectra and the values obtained from lamps 50 W 230 V HI-SPOT (Havells Sylvania.25a aw ± SD 0. which works algorithm of interval selection (Holland. Finlandia). The effect of storage time on the variables (TVB-N. after 14 days under the conditions of the experiment the samples were no longer fit for human consumption. R being reflectance.00b 0. doubling the initial content in 14 days (Table 2). Darmstadt.00c pH ± SD 5.986 to 0.44a 75. obtained for samples without packaging film were labeled “WPF” and All destructive analyses were done in triplicate after grinding and those with packaging film “PF”. 2007). De Baerdenmaeker.06b 6. Duplicate pour plates were prepared per dilution and values was evaluated using a discriminate analysis. Wallays. The position regression (PLS) (Bharati & Champagne.95 ± 0. The values of water activity in the chicken samples increased from 0.00a 0. Natick.2. 2).14 ± 0.1% peptone water were prepared from each log(cfu g− 1)) was subjected to a variance study (ANOVA). / Food Research International 44 (2011) 331–337 333 2.93 ± 0. The influence of packaging film and storage time on the spectra Germany). Zhang.05). Martens & Naes. & Huff-Lonergan. The results showed that there was a significant increase in TVB-N values with time (P b 0.3. 1989. R. Ntzimani.0. Alemania) and Specim ImSpector V10 1/2″ selected wavelengths were obtained from the IPLSDA genetic filter (Specim Spectral Imaging.1.. some authors Table 2 Means values of moisture (xw). water activity (aw).D. USA).06c Fig. 1975). this parameter can give information on biochemical changes occurring during product storage.05) with storage time. so breasts with standard deviations (S. These wavelengths are distributed at intervals of 0. The diffuse reflectance spectrum was collected using 1040 different 3. The influence of instrument was operated by the software pylon Viewer 2. Gardner.80a 20. a (n = 3) and 1 ml aliquots were plated in duplicate. 7 and 14 days). LTD.06a 5. In order to sample obtained from every container on every measurement day describe the relation between the variables and storage time. among others. France) the destructive analyses was obtained by partial least square producing an indirect light to reduce reflections (Fig. three quarters of the total samples were used for Images were taken using a CCD 102f camera (Basler Vision calibration and cross-validation and the remainder to test. which would reduce water retention capacity (Huff-Lonergan & Lonergan. Although TVB-N is not regulated for chicken meat. 7. Destructive analyses wavelengths (each wavelength is digitalized by 8 bits). Merlucii- dae and Gadidae.993 ± 0.5769 nm in the range The results of the destructive analyses of packed and sliced chicken from 400 to 1000 nm. 2006).993 after 14 days (P b 0. Missotten.95 ± 0. Means in a column with different letters are significantly different (P ≤ 0. Pournis. (Basler Vision Technologies. Similar values have been found by other authors (Economou. package (The Mathworks. Nondestructive analysis. Spectral acquisition setups. 2009) was employed to obtain predictive methods of 2.1. 1). The main Technologies. 2005. and 14 at 4 °C are that an image is grabbed with the resolution of 1392 × 1040 pixels. Although European legislation does not set microbi- ological limits. 1. 2009). Spectra data incubated at 28 °C for 48 h.987 during the first week and to 0.16 ± 0.05). Oulu. used. & Savvaidis. As the limit set for fish is 30 mg/100 g.48a 75. pH and TVB-N and their standard deviation (SD) of the packaged sliced chicken breasts during storage (0. This could have been due to the enzymatic action on proteins.987 ± 0. Germany). Partial least squares discriminant analysis (PLSDA) (Brereton.986 ± 0. . Time (days) 0 7 14 w x ± SD 74. Ahrensburg. Samples were selected at random from two groups. Gennevilliers.02b 33.48 ± 0. Ahrensburg.3. Lonergan. Spectra collection calibration and storage time.. The aerobic plate counts (APC) increased significantly during storage (Fig. 1992. shown in Table 2.) on days 0.04c TVB-N (mg/100 g) ± SD 15. TVB-N is a parameter regulated by the Economic European Community (EEC. like a linear multispectral camera. of the illuminant and camera to the sample was always constant to All statistical procedures were performed on the Matlab software control the lighting conditions and to obtain a constant image size.05). Statistical analysis The aerobic plate counts (expressed as cfu g− 1) were estimated following the method proposed by the UNE-EN ISO 4833:2003. Massachusetts.81 ± 1. K1 and Tenfold dilutions in 0. 2004).93 ± 0. Grau et al. Microbiological analysis 2.1. 2005) for some fish species such as Salmo salar and the Pleuronectidae family (except Greenland halibut). The distance between the illuminant and the sample was 18 cm. Results and discussion (scanned line) obtained was composed of 256 gray levels (8 bits). The image 3. The significant difference (P b 0.1664 sampling time was therefore evaluated. & Saeys. For all parameters except moisture (xw) there was a Reflectance data were stored as log (1/R). xw.39 ± 1. except for Salmonella (EEC. aw.

Three ATP derivatives (inosine 5′-monophosphate. Lee. 3 shows that K1 pH values continue decreasing for another 24 h. but that there had been little degradation of Ino to Hx. 2007). The first principal component (PC1) explained Values shown are the means of triplicate measurements. autolytic action of enzymes.7 μmol g− 1 at day 14. Ino analysis should be carried out to confirm this. reaching values of increased in value from 0 to 9 days of storage. Since the K1 value could be commercialized for a period between 9 and 12 days after proposed by Saito. Dyer. Broadhurst. pH and parameter K1. 2000. Ino concentrations were constant or even log (cfu g-1) showed a slight increase until day 7 of storage. According to these limits. as well as enzymatic The evolution of major adenine nucleotides and their related activity of microorganisms. 8 IMP content decreased with storage time. inosine. which excludes reached their highest values at the end of the study (Table 2). 9 100 8 Time = −695:432 + 592:514*aw + 0:104661*K1 ð2Þ 7 −1 80 + 0:0207147*TVB−N + 16:5139*pH + 1:21612*log cfu g 6 C (mmol/g) K1 (%) 5 60 2 R = 99:27% 4 3 40 3. 4 shows the Fig. Ko. 1983. Bars express standard deviation. after which a 6 considerable reduction in this compound was observed from this point to the end of storage. increasing rapidly throughout the second week of the study.51% of total variance. of freshness caused by autolytic changes. Fig. 12. although additional sensory hypoxanthine) were identified and quantified. 2000). breakdown products and due to the rapid disappearance of adenosine During the first 7 days pH values decreased significantly and phosphate. 2001). recommended different limits to classify meat quality.334 R. breed. with a R2 = 0. 3.7% of the . Dalgaard (2000) reported that the formation of IMP is due to the Fig.2. & Dingle. Ritvanen. and Hx concentrations during storage of packaged sliced chicken breasts Storage time fitting (Eq. while bacterial spoilage contributes to Ino 9. 3 by the end of the second week.993. which increases pH values. Clarke. Fig. & Song. when the IMP was depleted which could be correlated with the loss Ma. during the first 7 days. 73. The evolution of IMP. Large variations in nucleotide content in hence microbial numbers reach higher levels and raise TVB-N and K1 muscles have been found not only due to age. then remained 5. Alakomi. sensory quality is correlated with increases in Hx level during storage (Giménez. observed that the use of Hx content without any other Calderón. 1990). & Goodacre. 2. freshness decreases. log cfu g− 1.4 μmol g − 1.4 (Richardson & Mead. Greene & Bernatt-Byrne. reaching levels of 6. sliced chicken breasts Nowlan. In spite of the fact that it would explain packaged chicken meat spoilage. The images were therefore taken Hx IMP Ino K1 before and after the chicken samples were unpacked. Vainionpää & intrinsic enzyme activity and nucleotide concentrations (Hiltz. adenosine-5′-diphosphate (ADP) and the first 2–3 h after slaughter pH values decrease due to lactic acid adenosine-5′-monophosphate (AMP). 2004). This could imply an few days could be due to a reduction in electrolyte dissociation and an increase in nucleotide degradation. onstrated to be a good freshness index in fish. / Food Research International 44 (2011) 331–337 9 same species (Ellis. (1984) proposed K1 value. Arai. & Song. and 14. Karube et al. However. sex. inosine—Ino. These results are in agreement with the compounds can provide information on the biochemical changes evolution of microbiological counts (Fig. the Ino had been completely degraded 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 t (days) to Hx. The dotted line is the limit of and Hx formation. Values lower Botta (1995). other studies on chicken breast (Kavitha & Modi. (2)) showed a linear relationship between is shown in Fig. Kim. increase in ammonia formation (Chu. 12 and 14 days. 7. this relation is not used as a freshness indicator. The K1 value has been dem- production in red muscle. 2). although Chen et During the spoiling process of chicken meat. 2005). Grau et al. 2005). 1997). Initial IMP levels were 4. During adenosine-5′-triphosphate. Hx showed a slow increase during the first 5 week of storage. The increase recorded in the last practically constant until the end of the study. which would imply a large degree of spoilage during this week. Kavitha & Modi. 3. Yu. Roncalés. while PC2 explained only 19. These results indicate 4 that. occurring during storage and may well be correlated with meat freshness. 1971.6–5. High levels of Ino and low of Hx were found on day 0. and species levels as a result of converting amino acids to a growth substrate (Bailey. reaching very low levels 7 at days 9. 2007) but also between muscles of the instead of glucose. For white muscle. 0 2 4 6 8 10 12 14 The spectra obtained (log (1/R)) for each image were classified t (days) according to the sampling time (storage time) based on the presence or absence of the packaging film. unpackaged samples. 2009. Nondestructive analysis (image analysis) 2 20 1 An assessment of the effect of the addition of plastic film was 0 0 performed on diffuse reflectance data acquired during storage time. and samples should be less than 12 days. Ahvenainen. Aerobic plate counts (APC) for broiler chicken breasts stored at 4 °C during 0. part of the IMP had been transformed to Ino. 3. Nychas & Tassou. Changes in concentration of nucleotide compounds (inosine 50-monophosphate—I discriminate results performed on the data of the packaged and MP. working with fish processed during the latter stages of than 10− 6 cfu g− 1 would indicate good quality meat (Pascual & its storage life. and Matsuyoshi (1959) includes intermediate packaging. All these results may indicate that the marketing period for these as in fish. such as chicken breast. al. (2002) found higher levels. hypoxanthine—Hx) and K1 values (with standard deviation bars). & Beltrán. Several authors have also found that the loss of acceptable contamination. which agrees with time and TVB-N. reaching its maximum at day 9. APC counts between 10− 6 and 10− 7 cfu g− 1 would index of freshness may be misleading. The large variation observed in have acceptable quality and counts higher than 10− 7 cfu g− 1 would the rate of nucleotide degradation is probably caused by variation in be unacceptable (Smolander.

remove baseline shifts and increase apparent spectral resolution (Lin et al. Thus.E-04 1..65 when these were added to the study PF 7 -1 employing only the 13 selected points (R2 = 0. the slight influence of the packaging film was absorption bands at 545.1231 -6.8894 0. 7 and 14 days). the cross. 575 and 635 nm suggest a dynamic not taken into account for the subsequent studies. although the -10 derivatives are notoriously sensitive to noise (Tsai & Philpot. 7 days PF 0 information provided at the turning points of the second derivate of 2 the spectra at 525.. for all selected wavelengths. Since the aim of this study was to apply visual extends throughout the macrocyclic ring (e. 638. (Liu et al. The respectively). to 0.94).73%) PF 14 R2. Discriminate analysis for the spectra (log1/R) of the samples based on the SW-NIR spectrum with biochemical components that change during presence of packaging film (PF: with packaging film.9609 0. Liu et al.2236 R2 (C) 0. for PLS study..1193 0. since the differences between them were the greatest. This correlation is storage time. Charles.E-04 class 0 days 7 days 14 days 0 days 7 days 14 days -4. maximum differences between time 0 and 7 days were in the range of In order to increase the resolution. the root mean -5. or negligible. These that of the vibrating molecule. 2008. 2008. although wavelengths around 638 nm were influential too. The sliced chicken breasts. The second 14 days WPF 7 derivative technique is often used to process NIR data. 426. 460.E-04 time 0 days time 7 days SE (C) 0 0 0 0 0 0 SE (CV) 0. 480. 2003). who found that the energy versus wavelength called a spectrum (Osborne. Some authors have correlated some wavelengths of the visible and Fig.75 obtained. The low levels of red pigments (myoglobin Table 3 shows the statistical results of the partial least squares derivatives) in chicken breasts could be responsible for the changes discriminant analysis (PLSDA) employed to obtain the predictive observed at 638 nm. The addition of film to meat samples caused only slight.9359 0. 946. The prediction results for an independent test set are expressed as both the standard error (SE). 5 shows the second derivate of the spectra for all storage times -4 WPF 14 and the selected wavelengths using the IPLSDA algorithm. 970 and 14 days. 2004).g.0781 0 0 0 -5. -10 -5 0 5 10 15 20 Most selected wavelengths were in the range 400–500 and 940– PC1 (73. -2.82) obtained at day 0 are shown in Fig. The results of the cross and test set validations performed for the predictive models employing the values obtained for the selected 2. the haem prosthetic and SW-NIR techniques to monitoring the freshness of packaged group in myoglobin Mb) (Andrés et al. root mean square error (RMSE) and the coefficient of identified at 980 nm (OH second overtone) (Andrés et al. It helps to -7 separate overlapping absorption bands. .1137 0. characteristic bands of water were error (SE).E-05400 500 600 700 800 900 1000 square error (RMSE) and the determination coefficient (R2) for calibration (C) and cross-validation (CV) model. The greatest differences were and was slightly lower for 0 and 14 days (R2 = 0.9352 0. When the frequency of the radiation matches slight difference between the two groups at day 14 of storage. 2003). selecting the main spectrum points between samples at 0 days and days 7 and 14.8673 0. but not between days 7 with the best correlation with storage time.7179 0. after 499 nm could be responsible for discriminating between 7 and were 413. 1998).9513 0. 499. A weak near-infrared band at 760 nm could be methods of classification according to storage time.1094 0 0. The calibration due to the OH third overtone or an absorption band produced by (C) and cross validation (CV) accuracy was given by the standard myoglobin oxidation.E-05 Statistical results of the square regression (PLS). there was a interatomic distance. employing the variable selection algorithm of IPLSDA.. considering all conversion and degradation for a number of myoglobin derivatives samples (with and without packaging film) to be equal. & Tormod. Grau et al. 5) reduced the value of PC2 (19. 5.8754 0. Moreover.72 and R2 = 0. WPF: without packaging film) and meat storage (Andrés et al.. 2003). / Food Research International 44 (2011) 331–337 335 8 wavelengths were better than those obtained employing the entire spectrum (Table 3). 2000).1110 0. R.1666 0. Although the samples could not be differentiated based on there is a parabolic relationship between potential energy and the presence or absence of plastic film at 0 and 7 days. 2008.0933 0.89). The wavelengths and 14 (lowest difference). 982 nm.1580 0. 2003). Second derivative of the spectra for all storage times. 6. Liu determination (R2). Selected wavelengths before 460 nm and obtained.1040 0. (◊) points selected by R2 (CV) 0.1987 0. 473.9419 0.. 449. 967.E-04 Table 3 5. calibration (R2C) was higher for all the times (R2 = 0.2984 0. wavelengths 430–435 nm form the Soret band attributed to an increases in the optimal prediction errors of the NIR and NIT methods..E-04 RMSE (CV) 0. WPF 0 Fig. related mainly to the vibrational frequency which is a function of the masses of the two atoms m1 and m2 and the strength of the bond k and variation.7766 IPLSDA.51%) 980 nm. 4. 570 and 600 nm (Fig. between time 0 and 14 days. 1992).E-04 Wavelength (nm) Modelled Entire spectrum Thirteen selected points -3. Liu et al. Differences between the spectra for the sampling time with those validation coefficient (R2CV) was only higher for 7 days (R2 = 0.7510 0. there was a clear discrimination based on storage (macrocyclic tetrapyrrole) rings in which electron delocalisation time (0.E-04 time 14 days Selected points RMSE (C) 0. This wavelength range could mainly differentiate variables available were reduced. intense π→π* transition observed in all conjugated porphyrin On the other hand.. 942. The information provided by the whole spectrum 0 days 5 could obscure that contributed by the selected points.9318 Fig. Although the determination coefficient for et al. the numbers of independent 460 and 499 nm. 560. there will be a net transfer of energy observations were consistent with the results observed by other from the radiation to the molecule which can be measured as a plot of authors (Isaksson.2878 0.

H. A.. W. Adenosine triphosphate catabolites as flavour compounds and freshness indicators in Pacific cod (Gadus macrocephalus) xw aw pH TVB-N (mg/100 g) K1 log (cfu g− 1) and pollock (Theragra chalcogranama). B. Factors affecting the with the physicochemical.0693 3. J. Evolution of volatile fraction and herein is fast. (2008)... 460.. International Journal of Food Microbiology. Commission Regulation (EC) No 1441/2007 of 5 December 2007. I. Gil.). (1975). Aristoy. pH. S. Lee B. aw and cfu g− 1.767 0. Journal of the Association of Official Analytical Chemists.. Silva. Li. J. Lin et al. L. A. F. F. 1470−1476. & Serra. J. In G. EEC (2005).336 R. & Goodacre.. Chapter I: Total volatile basic nitrogen (TVB-N) limit values mainly been associated with an increase in the values of TVB-N.. Different packaging conditions to Table 4 improve shelf-life of filleted rainbow trout. low-cost and nondestructive and we consider it suitable ATP related compounds during storage of desalted cod (Gadus morhua). Martins.756 0. spoilage of muscle foods: Current status and future trends. . Grau et al.0009 0. (1971). 499. The use of visible and near infrared reflectance spectroscopy to predict beef -0. J.0834 6. Gudjónsson. & Kee. C. Rapid and from the images. 19. will be carried out in due course. 217−224. indicating a Chun H. Development of a rapid method based on 2004. & M.). & Song K. doi:10. Y. Adaptation in Natural Selection and Artificial Systems. Of special interest is the linear correlation between quantitative detection of the microbial spoilage of meat by Fourier transform infrared the visible and SW-NIR spectroscopy and parameters such as the K1 spectroscopy and machine learning.. & Stéphane..0118 6. (1995). Broadhurst. B. J. Meat Science. Food Research International. & Champagne. Meat Science. (2008). Food Control. D. García-Breijo. Dalgaard P. J. The visible and SW-NIR spectroscopy method we report Technology.352 0. & Toldrá. B. Effect of UV-C irradiation on relation among these five parameters.. Nisin–EDTA treatments and modified atmosphere packaging to increase fresh chicken meat 4. NY: VCH physicochemical. 43−50). Washington. 51. TVB-N. Trends in Food Science & xanthine).. 14(5).). 53−59. 0. (2005). (2003). Faeroe Islands: independent test set are expressed as both the root mean square error (RMSE) and F∅roya Fródskaparfelag. Rowland. Tórshavn. (1985). (2001). & Beltrán. respectively. pH.. C. M. Clarke. R. 27–59. G. A. Food Control. / Food Research International 44 (2011) 331–337 0. C. Fernández-Segovia. New York... Greene..893 0. Bharati. Using near-infrared multivariate image regression to predict pulp properties. E. 415−423. Armero. variation of the spectrum at the selected point during storage time 444−448. R. J. Broadhurst. M. Economou. L 322. Qualitative and quantitative characterization of spoilage bacteria connection between spectral changes in meat and fish samples and a from packed fish.1016/j.. Statistical results of the square regression (PLS). 26. The University of Michigan Press. Fuentes.. L. 78. R. Barat. Fig.02 Wavelength (nm) (2008). H. I... G. (1983). Correlation between image analysis (nondestructive analysis) and the Botta. A. H. Bailey. J. E.. (1999).06. Applied and Environmental Microbiology. Pournis. E. Aristoy. 942. DC: American Chemicals Society. & Kanea. M. Laying down implementing measures for certain products under Regulation 2074/2005. (2004).3. I.03 0. J. A. 426... Science and Technology International. the determination coefficient (R2) for calibration (C) and predictive (P) models. N. 638. The loss of freshness in sliced and packaged chicken breast has L 338. 967. 191−195).789 and quality control (pp. 37−47.. aw and the microbial loads (cfu g− 1). B. 3. Nowlan.0016 0. Niclasen (Eds. T. -0. 257−258. Escriche.. M. Fernández-Segovia. Wu. Kell. Section II. In A. inosine and hypo. 130. P. 2002. Multivariate pattern recognition. The presence of packaging film did not affect the information obtained Ellis. biochemical and microbiological para- inosine monophosphate content of muscles in Taihe silkies chickens. (1995).4210 0. Seregély. Sensory characteristics of cooked pork loin as affected by nucleotide content and post-mortem meat quality. E. R. J. M.07 Difference (time 0 -time 7 days) Acknowledgements Difference (time 0 -time 14 days) 0. Andrássy. Martínez-Máñez. 970 and 982 nm. B. reliability and accuracy studies. & Dai. G. (2008).. (2002). Soares-Pereira. Similar behaviors were found the inactivation of inoculated pathogens and quality of chicken breasts during storage. K. S. & Dingle. R. (1992). J. 36. D. & Goodacre. J.01 0 Alain. Commission Regulation (EC) No 2074/2005 of 5 December 2005.593 0. Kim J.1478 0.02 References 0. I.04 the PAID-06-08-3251 and GVPRE/2008/170 Projects. The prediction results for an Proceedings of 30th WEFTA Plenary Meeting (pp. Freshness.. Brereton (Ed.5294 0. Ntzimani. Differences between spectra at 7 and 14 days to those obtained at day 0 for the Feather (Eds.. I.804 Michigan... which indicates the simultaneous evolution of Ellis. 55. Official Journal of the European Union visible and SW-NIR spectroscopy have been related to storage time of 1441/2007. I. & Serra. 319−333. 2008. Toldrá. Study of sea bass (Dicentrarchus labrax L.. & Bernatt-Byrne. A. 473. Fish inspection R2 (C) 0. S. front-face fluorescence spectroscopy for the monitoring of fish freshness. (R2P) was obtained for K1. J.. decrease of freshness associated with microbial growth (Ellis et al. M.2009. & Savvaidis. S. (2009). Conclusions shelf-life. D.. The Maillard reaction and meat flavour. I. D. D. F. analysis) Brereton. Journal of Food Science. M. Horváth. M. 413. (1990). Andrés.7264 Hiltz. G. Bruno-Soares. for certain categories of fishery products and analysis methods to be used. Chen. biochemical and microbiological analyses (destructive Publishers Inc. and TVB-N indexes. 480. longissimus thoracis et lumborum quality attributes. 12. Chemometrics illustrated by case studies..799 0. (2009).8190 4. 8−14. The spectra Journal of the European Union L 338/36.01 413 426 449 460 473 480 499 638 942 946 967 970 982 87(2). (2008).. Food Chemistry. R2 (P) 0..7994 indices by biological and technological factors. (2002). Food for monitoring the freshness of sliced and packaged chicken breast. 414−424. J.006. F-FE 380A/00.. London: Fishing News (Books) Limited.736 0. Dalmadi & Farkas. 449. Evaluation of seafood freshness quality (pp. 3(5). J. Roncalés.... the samples.. Food Chemistry. G. Z. R. EEC (2007). Barat. Li. The highest determination coefficient for prediction Australian Journal of Animal Science. Asian meters (Table 4). A.). Effect of superheated steam on the inactivation of Listeria innocua surface-inoculated onto chicken skin. 946. 15(9). 68. Walter. Further repeatability. K. 108.. 169−184). 2822−2828. D. M. Holland.. Xie.878 0. P.. Journal of Food Engineering. et al.Official parameter K1. X. Tappi Journal. 757−763. D.. 1359−1363. 681−688. S. 6. 12–29. Burns.883 0.. 1648−1654. RMSE (C) 0.03 M. 180).05 We wish to thank the Polytechnic University of Valencia and Generalitat Valenciana for the financial support they provided through Difference (log 1/R) 0. 2006). In R. Dufour. (2000). & Goodacre. H. quality and safety in seafoods.580 0. N. R. E. D. Kreuzer (Ed.. & Murray. P. H. Dyer. selected wavelength. In R. Frencia. F. Ann Arbor.. Amsterdam: Elsevier. Analyst. D. C. Rapid and quantitative detection of the microbial certain metabolites (inosine 5′-monophosphate. E. packaging methods. P. -0. obtained from sliced and packaged chicken breast images using Microbiological criteria for foodstuffs.foodcont. I... & O. Giménez. 162−171. J. Freshness monitoring of sea bream (Sparus aurata) with a potentiometric sensor. A. 114. Rapid identification of closely related muscle foods by vibrational spectroscopy and machine learning. Variation of biochemical quality RMSE (P) 0. using other Flores.06 0.. R.. Yu D.) salting process: Kinetic and thermodynamic control. (2000).3130 0.. where the best wavelengths for modeling freshness were Ellis. Liquid chromatographic determination of hypoxanthine A partial least square regression (PLS) was performed to detect the content in fish tissue. The Maillard reaction in foods and nutrition (pp. Flair-Flow Europe Technical Manual. 68. P. by other authors who worked on larger wavelengths and observed a Dalgaard. I.

M. 41−51. H.. M. W. M. fat and protein content of raw meat by NIR. Jin. Spoilage processes and proteolysis in chicken as study using near infrared spectroscopy to evaluate freshness and detect spoilage in detected by HPLC. & Saeys. (1998). E. 343−353.. 36. M. Encyclopedia of analytical chemistry: Applications.. N. 67−73.. Williams. Cavinato. 749−752. M. Ch. 293−300. Journal of Food Science and Nutrition. Watanabe.. 38–39(4). C. N. Mousavi. pork freshness using potentiometric sensor. S. Pringle. 2003). 1107−1115. R. G.. (Eds.).. 12.. K. A. Near-infrared spectroscopy in food analysis. Meat Science. C... B. Windham. M. In Robert A. B. 62.. Alakomi. K. Letters in Applied of Environment. Lyon. C. J. & Lonergan. Realini.. Journal of Bionic Engineering... Sinelli. Minn: American Association of Cereal Chemists. 119−130. Andrássy. Meat volume 12: Electronic absorption and luminescence. Osborne. & Namsoo. 24. Applied Spectroscopy. N. B. V. Determination of Sinelli.. 148−155. (1989). A time–temperature indicators as Ko. K. Gardner. (2000). H. 27−32. & Yu-bin. Meat Science. 66.. B. & Feng. 1169−1172. N. Zs... Acribia. Effect of water activity and temperature on Smolander. Remote Sensing breasts by diffused reflectance spectroscopy (600–1100 nm). P. kinetic determinations (pp.. H. stored in different temperature conditions. (1989). H. H. 194−204. 24.. Tanaka. & Takimoto. . B. packed industrial ricotta cheese. 71... C. Dong-hui. (1996). Multivariate calibration. A. Acta Zhang. Suzuki. 32(1). & Huff-Lonergan.. St. 74. M. Giardina.. 5. major compounds in milk powder and wavelength assignment. (2000). 245−254. B. J. 13. 74. Lin. A. J. .. Effect of chlorine dioxide treatment on microbial quality-indicating tools. (2000). (1978). Biochemical basis of post-mortem 122−129. Y. Charles. & Duckett. I. Food Control. predicting moisture. 217−229. (2007). V. Al-Holy. He. F. G. & Tao. M.. G.. nucleotide catabolism in cod (Gadus morhua) and its relationship to spoilage. K. / Food Research International 44 (2011) 331–337 337 Horváth. CA: Varian Associates. degradation of 5′-inosine monophosphate in a meat model system..technique. (2004). (2006). B. E. (2006). A. E. K. F. Journal of Food Science. Journal of Near Infrared Spectroscopy. 50(9). and water in plastic-wrapped. Ch. K. 1685−1694.. Journal Food Science... J.. T. & Leblanc. & Callis. texture. De Baerdenmaeker. (1983). Derivative analysis of hyperspectral data... H. H. New York: John Wiley & Sons Inc. E. Johnson. K. 42−44..). Colony-count infrared spectroscopic method for the detection of spoilage in rainbow trout technique at 30 degrees C. B. G. Quantitative analysis in basic liquid Zaragoza. C. 93−102. & Tassou.. theory and instrumentation meat: The role of postmortem biochemical and structural changes (Review). Applied Spectroscopy. gas chromatography. M. S. I. & Nielsen. & Hotori.... T. T. Richardson R. A.. Horizontal method for the enumeration of microorganisms. C. S. (2008). Analytica Chimica Nádai. Journal of Agricultural determination of protein. Missotten. D. 65. Characterization and meat freshness Pascual. 217−221. (2006). M. M. matrix for detecting beef freshness. A preliminary Nychas. (1984). Talanta. R.. P. 75−81.A. Y. Electronic nose with an air sensor Alimentaria. 1280−1286. R. P. C. P. E. J. New Food. Journal of food of postmortem changes of integrin. Watanabe. moisture.. 47. Rapid and quantitative detection of the microbial spoilage of chicken Tsai. Gill. (2005). T. Vainionpää. Journal of Food Mateo... & Tormod. (2005). Determination of Liu. 40.. Journal of Agricultural and Food Chemistry. Microbiology. Prediction of color. Saito.. T. Journal of Agricultural and Food using Fourier transform near infrared spectroscopy to monitor the shelf-life of Chemistry. 199−208. I. Mechanisms of water-holding capacity of (Ed. Tsuneishi. sliced pork meat. T. K.. E. M. C. & Modi. D. Aguirrezábal. Yong-Jin. E. steam distillation. Ito. 1−14). 104(1).. M. 13. (2002). 404.. 223). V. (1984). Near-infrared technology in the agricultural and food 43. Seregély. E. & Calderón. 39. S18−S23.. Short-wave near-infrared spectroscopy analysis of infrared reflection technique.. compounds. T. 389−403. Hansen Way. Martens. A new method for estimating the freshness Kaneki. 37(1). S.. Kavitha. (2009). Meyers Huff-Lonergan. S. (2001). & Matsuyoshi. (1999). Paul. 232−242. International Standards Organisation. A. G. Visible/Near infrared Zhang. LWT .. Bulletin of the Japanese Society of Scientific Fisheries.. M. Sensory quality criteria for five fish species Martens.. & Farkas. industries (2nd ed. Mousavy-Hesary. & Rasco. 15.. Rapid quantitative determination of trimethylamine using Engineering. Heia. predicted from near infrared (NIR) reflectance measurement.. A new tool for the evaluation of fish freshness. 314−319. K. (1959). E. M. Arai. Naczk. infrared Science.. Lonergan. Brian G. 1821−1826. waveband selection for on-line measurement of grain cleanness. J. Matsuoka. & Mihályi-Kengyel. spectroscopy. P. M. and sensory characteristics of beef steaks by visible and near Wallays. 1−7. Hyperspectral infrared reflectance spectroscopy. & Cattaneo.. Al-Holy. Meat Science. (1989). Y... (1988). (2003). Monitoring the shelf life of dairy products. W. growth and qualities of chicken breast. & Ahvenainen. 71(1). Biosystems Malle. Dalmadi. Wiley and Sons. (2001). Madrid: Díaz de Santos S. & Pink. L. Warm. Shimada. D. A photodiode-array-based near-infrared products in beef by reversed-phase HPLC. & Philpot. D. Nondestructive NIR and NIT hake: Use of Fourier transform near-infrared spectroscopy. Journal of Food Protection. (1987). P. Grau et al.. Analytica Chimica Acta. Ritvanen. Investigations of different equations Acta. T. & Toyama.. UNE-EN ISO R1442 (1979).. Al-Qadiri. M. A feasibility study. Domínguez. 67. & Naes. & Norris. 54. & Sigernes. J. 10(2).. H.. T. T. S. 610(2). (2005). 44(4). A preliminary study fish freshness with an enzyme sensor system. Taste Quality. K.. & Rasco. fat. UNE-EN ISO 4833:2003 standard (AENOR. (2001). Z. Contribution spectroscopy. Chichester.. A. (2008). J. Nilsen. Ciencia de la Carne de Ave. M. & Stevenson. J. (2008).. E.. Cavinato. Ma. (1997). Palo Alto. M. D. Microbiología alimentaria: Metodología analítica application of a serial three-enzyme reactor system measuring ATP-degradative para alimentos y bebidas. (1992). (Oncorhynchus mykiss) fillet. Journal of the Science of Food and Agriculture. Nádai.. J. Measurement of of fish. Acta Alimentaria.. homogenized meat. Miura. A. W.. and Mead G. S. M. chromatography (pp. Preliminary experiments for measuring meat composition by near Wu. spectrophotometer for the 600–1100 nm wavelength region. A. (2004). W. 578−585. S. S. Analysis of ATP and its breakdown Mayes. Rapid near stuffs. In-Seon.. M. holding capacity of pork. Acta Alimentaria. desmin and μ-calpain to variation in water Science.. K.). T. and Food Chemistry. Esaiassen... Pink.. (2004). S. & Song. 4280−4284. E. Surette. Karube. Y. Microbiology of food and animal feeding Lin. R. H. 19−22. Evaluation of the quality of frozen minced red Isaksson. Barzaghi. & Zumalacárregui. 756−760. 46(11). UK: J. compounds in chorizo and their changes during ripening. N. K.Food Monitoring of the quality of modified atmosphere packaged broiler chicken cuts Science and Technology.